This project investigates effective and resource efficient establishment of connectivity among disjoint wireless sensor network (WSN) segments. The segments can belong to a structurally damaged network caused by the failure of multiple sensor nodes. In addition, the segments can simply be standalone WSNs that are normally operated by different agencies and are to be federated to serve a common application. The objectives of this project are to develop novel solutions for various aspects and contexts of the federation problems, to create a prototype for validation and to share the results/experience with application designers.
The technical approaches consider the availability of resources such as mobile sensors, mobile and static gateways and their count. Both optimal and heuristic solutions for repositioning of mobile sensors and placement of mobile gateways are studied to establish connectivity as well as achieving some desired performance (i.e., QoS). Finally, the results are validated via a real test-bed consisting of sensors and mobile robots. This project will boost the effectiveness of many civil and scientific applications. Example of such applications include crisis management, where existing WSNs may suffer an extensive loss of nodes due to fire, flooding, debris, etc., or when the services of networks owned or controlled by different parties or agencies need to be aggregated to assess in search-and-rescue. The results are made available in various forms including archival publications, tutorials and web-based resources. The project is enriching the curricula at UMBC and SIUC through hands-on projects and attracting K-12 students via prototype demonstrations.
Wireless sensor networks (WSN) can increase the efficiency of many real-life applications through the collaboration of hundreds of miniaturized sensors which can be deployed unattended in harsh and inhospitable environments. While such deployment eliminates/reduces human intervention and provides fully-automated data gathering systems, WSNs are still prone to sensors failure which not only can degrade the quality of coverage but also disrupt the data traffic. To address such a problem, most approaches in the literature deploy redundant nodes during network setup and reconfigure the network topology to establish alternate data paths. However, sometimes the network suffers a large scale damage involving many nodes and gets partitioned into multiple disjoint segments. For these cases, a provisioned approach for tolerating occasional failures at the network design level will not be effective. A similar scenario is when an application can be leveraged by engaging a number of standalone WSNs. For example, the disaster response may involve collaboration among multiple government agencies and may need to combine the services of their autonomously-operating WSNs in the search-and-rescue process. The main goal of the project is to quickly repair a structurally damaged WSN or linking multiple disjoint WSNs with the least resources and overhead. This process is called a federation. Federation of multiple network segments is characterized as short lived and rapid in order to ensure responsiveness to application needs until a more stable solution is developed or the need for combining the network services ends. Repositioning/deployment of mobile nodes is the main methodology pursued in the project for achieving the desired federation Throughout the funding period, major contribution has been made both on the research and educational fronts. The conducted research has led to a body of knowledge upon which the designers of WSNs can rely. The technical outcomes of the project include: (1) developing algorithms for distributed partition detection and optimized node repositioning for restoring connectivity while increasing responsiveness; (2) developing schemes for estimating the minimal count of additional mobile nodes which the WSN would need in order to restore connectivity while achieving some desired performance. Again, this applies to both partitions of a partially damaged WSN and to a set of standalone WSNs; and (3) building a prototype system for validating the proposed algorithms and providing hands-on experience for graduate and undergraduate students. These outcomes can boost the effectiveness of many civil and scientific applications of a national and international interest. Examples of these applications include crisis management, where existing WSNs may suffer an extensive loss of nodes due to fire, flooding, debris, etc., or when the services of networks owned or controlled by different parties or agencies need to be aggregated to assess in search-and-rescue. Similar scenarios can be identified in border protection and combat zone reconnaissance where the damage may be caused by bombing or sabotage. The research results also apply to space exploration where assets are scarce and federating the services of previously deployed segments of networked sensors would need to be engineered. In summary, a comprehensive library of recovery algorithms has been developed to cater for various application requirements. The results of the investigation have been published in numerous papers in refereed conferences and archival journals On the educational front, a new course has been introduced to the graduate computer science and computer engineering programs at UMBC. The course is gaining popularity among the students. When offered in fall 2010 and spring 2012, the course was filled to capacity. In the fall 2013 semester the course capacity had to be grown in order to accommodate the increased demand. Five Ph.D. dissertations and numerous MS theses have been and are being pursued on topics related to the project. A workshop was organized and numerous presentations and invited talks were delivered to raise awareness, share technical insights and stimulate interest. To further share the gained expertise, a book chapter has been published to highlight the issues associated with topology repair and point out appropriate solution strategies. The scope is further extended and published as a survey paper that can be referenced by practitioners and students exploring this area of research.
